[Insight-users] How to compute the local maximas in an image

[Oliver Trebbe]

Gaëtan Lehmann schrieb:
>
> Le 19 déc. 08 à 19:43, Oliver Trebbe a écrit :
>
>> Oliver Trebbe schrieb:
>>> Gaëtan Lehmann schrieb:
>>>>
>>>> Le 19 déc. 08 à 15:29, Oliver Trebbe a écrit :
>>>>
>>>>> Hello again Guys :)
>>>>>
>>>>> i just need the number of the maximas, where they are can also be 
>>>>> important but not necessary...
>>>>
>>>>
>>>> Hi,
>>>>
>>>> You may want to try an itk::RegionalMaximaImageFilter followed by 
>>>> an itk::ConnectedComponentImageFilter. Once the filters updated, 
>>>> you can get the number of regional maxima with the method 
>>>> itk::ConnectedComponentImageFilter::GetObjectCount().
>>>>
>>>> Maybe regional extrema is not exactly what you want though.
>>>>
>>>> Regards,
>>>>
>>>> Gaëtan
>>>>
>>> Hi,
>>>
>>> dont know where´s my failure, but it says there is no 
>>> itkRegionalMaximaImageFile header...
>>> so i used the Valued, this one works...
>
> both are in the Review directory:
>
> [glehmann at pixel2 Insight]$ find Code/ -name 
> itk*RegionalMaximaImageFilter*
> Code/Review/itkRegionalMaximaImageFilter.h
> Code/Review/itkRegionalMaximaImageFilter.txx
> Code/Review/itkValuedRegionalMaximaImageFilter.h
>
>
>>>
>>> cause of "Maybe regional extrema is not exactly what you want though."
>>> I want the 1 derivate of my 3D Image with extracted Regions (not in 
>>> a ROI just in a Image in a List) to know how many watersheds i need 
>>> to use or lay above it, or how to say...
>>> (maybe im also just wrong in my thoughts and u know how to do it 
>>> better :))
>>>
>> And to explain it better:
>> My Image is (imho) defined as a 'model' of a Polynomial of n degrees 
>> so i want a Polyonomial to fit into it, and then i want to get the 
>> Polynomials extremas...
>>
>
>
> From your last explanation, I still think that the regional maxima 
> filter followed by the connected component filter would give you what 
> you want.
> But I don't understand the previous one, with the watershed. Maybe it 
> would be easier with an image?
>
> Gaëtan
>
No Images here right now :(
Not in the Department anymore...
So maybe i can explain the image or my problem:
I´ve masked a FLAIR Image to get just the White Matter,
this was segmented so i get Regions with Lesions in it, but
some Regions include more than 1 Lesion and for this Regions
I need the Maximas to 'watershed' them and split them up...


-- 
Oliver Trebbe
Department of Neurology
University of Muenster
48129 Muenster, Germany